Predict the weekly sales of walmart stores by section using historical sales, markdown data and macro-economic indicators

Setting the working directory

Reading the required data files

Exploratory Data Analysis (EDA)

now we wanna see if this is a systematic issue or random

Since it affects a small percentage of the data, lets drop these records

Creating a function that does all of the above tasks in one go

get_basic_stats(stores_data)

Converting variable to correct datatype

Missing Value Imputation

Markdown variables are missing in over half of the cases, and the most logical reason for that seems to be the absence of any markdown in those cases

Data preparation

Stores unique key = Store

Features Data - Store+Date

Store+Dept+Date

Merging datasets

Dividing data into train and test

getting the date variable in the correct format

match the number of weeks of difference with the number of distinct weeks in the data

Getting the date that comes at 70% of the data and using it to divide the original data

Univariate Analysis

Bivariate Analysis

Variable trend with Dependent Variable

Date trends

Feature Creation

Modeling

Prediction 1

Prediction 2 - Random Forest Regression

Prediction 3 - ARIMA, another statistical technique

Model Comparison

Discussion and Conclusion